OpenLoop Research

AI research assistant that searches the web and synthesizes comprehensive answers. Built with LangGraph, OpenAI, and real-time web search.

How it works

1. Takes a question and generates optimized search queries
2. Searches the web using Tavily API
3. Reflects on information gaps and searches more if needed
4. Synthesizes final answer with proper citations

Quick Start

Prerequisites

• Node.js 18+ and npm
• Python 3.11+
• OpenAI API key
• Tavily API key (required, for web search)

Setup

Backend:

cd backend
pip install -e .
echo "OPENAI_API_KEY=your_key" > .env
echo "TAVILY_API_KEY=your_key" >> .env
python -m agent.app

Frontend:

cd frontend
npm install
npm run dev

Access at http://localhost:5173

Configuration

Set environment variables in backend/.env:

OPENAI_API_KEY=your_openai_api_key
TAVILY_API_KEY=your_tavily_api_key

Docker Deployment

./deploy.sh

Access at http://localhost:3000

Stack

• Backend: FastAPI + LangGraph + OpenAI + Tavily
• Frontend: React + TypeScript + Vite + TailwindCSS

API Usage

REST endpoint:

import requests

response = requests.post("http://localhost:2024/research", json={
    "query": "Your question here",
    "max_research_loops": 3,
    "reasoning_model": "gpt-4o-mini"
})

result = response.json()
print(result['answer'])

License

MIT